20 comments

  • ed 11 days ago
    Elegant architecture, trained from scratch, excels at image editing. This looks very interesting!

    From https://arxiv.org/html/2409.11340v1

    > Unlike popular diffusion models, OmniGen features a very concise structure, comprising only two main components: a VAE and a transformer model, without any additional encoders.

    > OmniGen supports arbitrarily interleaved text and image inputs as conditions to guide image generation, rather than text-only or image-only conditions.

    > Additionally, we incorporate several classic computer vision tasks such as human pose estimation, edge detection, and image deblurring, thereby extending the model’s capability boundaries and enhancing its proficiency in complex image generation tasks.

    This enables prompts for edits like: "|image_1| Put a smile face on the note." or "The canny edge of the generated picture should look like: |image_1|"

    > To train a robust unified model, we construct the first large-scale unified image generation dataset X2I, which unifies various tasks into one format.

    • nairoz 11 days ago
      > trained from scratch

      Not exactly. They mention starting from the VAE from Stable Diffusion XL and the Transformer from Phi3.

      Looks like these LLMs can really be used for anything

      • yieldcrv 11 days ago
        Pretty cool, comfy ui and community is too cumbersome for me and still results in too much throwaway content
  • lelandfe 11 days ago
    I left all the defaults as is, uploaded a small image, typed in "cafe," and 15 minutes later I am still waiting on this finishing.
  • bob_1200 11 days ago
  • ilaksh 11 days ago
    I think this type of capability will make a lot of image generation stuff obsolete eventually. In a year or two, 75%+ of what people do with ComfyUI workflows might be built into models.
    • freilanzer 10 days ago
      Well, at the moment it seems it's not working at all.
  • sswz 11 days ago
    Using a single model to unify all image generation tasks, including many computer vision tasks and visual language reasoning, could transform future image generation models. Although some capabilities, like text-to-image, aren't perfect, it's a significant advancement. The model's ability to integrate so many tasks with strong instruction-following skills is impressive. I'm excited about the broad impact OmniGen could have on future research.
  • block_dagger 11 days ago
    This looks promising. I love how you can reference uploaded images with markup - this is exactly what the field needs more of. After spending the last two weeks generating thousands of album cover images using DALL-E and being generally disappointed with the results (especially with the variations feature of DALL-E 2), I'm excited to give this a try.
  • 101008 11 days ago
    I am working on a API to generate avatars/profile pics based on a prompt. I tried looking for train my own model bt I think it's a titanic task and impossible to do it myself. Is my best solution use an external API and then crop the face for what was generated?
    • ncoronges 11 days ago
      The simplest commercial product for finetuning your own model is probably Adobe firefly, although there’s no API access support yet. But there are cheap and only slightly more involved options like Replicate or Civit.ai. Replicate has solid API support.

      Check out:

      https://replicate.com/blog/fine-tune-flux

      • 101008 11 days ago
        Is it Flux 1 possible to download and deploy to my own server? (And make a simple API on top of it?) I don't need fine tuning.
        • spaceman_2020 10 days ago
          The easiest flux api I’ve seen is with Fal.ai

          It is expensive though- Flux dev images are like $0.035/image

        • handfuloflight 11 days ago
          If you have GPUs on your server that can handle it.
    • haccount 10 days ago
      You can use a few controlnet templates and then whatever model you like and consistently get the posture correct. The diffusion plugin for Krita is a great playground for exploring this.
  • KerryJones 11 days ago
    Love this idea -- you have a typo in tools "Satble Diffusion"
  • gremlinsinc 10 days ago
    Anyone know how it handles Text? That's kind of my deal breaker, I like Ideogram for it's ability to do really cool fonts, etc.
  • wwwtyro 11 days ago
    With consistent representation of characters, are we now on the precipice of a Cambrian explosion of manga/graphic novels/comics?
    • Multicomp 11 days ago
      I sure hope so - at the very least I will use it for tabletop illustrations instead of having to describe a party's scenario result - I can give them a character-accurate image showing their success (or epic lack thereof).
    • jowday 10 days ago
      It’s not really consistent - or anymore consistent than, say, SDXL with IP adapter. Even in their example images the character they’ve input comes out wearing different clothes.
    • haccount 10 days ago
      I would say we already had one of those. There's more hand crafted human made content available than anyone cares to read.

      While this will enable a certain degree of more spam it will more importantly, on the positive side of things, democratize the creative process to those who want to tell a story in images but lack the skill and resources to churn it out traditionally.

    • fullstackwife 11 days ago
      not yet, still can't generate transparent images
  • empath75 11 days ago
    it seems like there's a lot of potential for abuse if you can get it to generate ai images of real people reliably.
    • hnbad 10 days ago
      We literally already had AI fake porn of Taylor Swift making the rounds a while ago. Prepare for women in public positions to face that kind of bullshit more frequently.
      • CamperBob2 10 days ago
        Eh, once it's ubiquitous, nobody will care.
        • cubefox 10 days ago
          Once fakes in politics are ubiquitous, people will stop trusting the real evidence.
          • CamperBob2 9 days ago
            That appears to have already happened, no AI required.
            • cubefox 9 days ago
              The trust in video evidence certainly can be much lower than it is now.
              • CamperBob2 9 days ago
                It's more an issue of indifference than trust. For instance, you can show Trump supporters any number of legitimate videos that depict Trump and his associates saying, doing, and promising all kinds of outrageous, offensive, and destructive things, and they won't care in the slightest. It's not that they don't trust the video, it's that they've been programmed not to care. The leader cannot fail.

                That's the ultimate purpose of disinformation -- it's not to make you believe false things, it's to make you believe nothing.

                So yes, AI fakery will contribute to that phenomenon on behalf of numerous bad actors, but it was always going to happen anyway. You don't need Hinton and Sutskever on your side if you have Aisles and Murdoch.

                • cubefox 8 days ago
                  > So yes, AI fakery will contribute to that phenomenon on behalf of numerous bad actors, but it was always going to happen anyway.

                  That's like saying: "Yes, crime might increase, but we will always have crime anyway." What will happen anyway is irrelevant precisely because it happens anyway. What's relevant is the expected increase in media distrust once everything might be a fake.

  • oatsandsugar 11 days ago
    I mean, I struggle even getting Dall-E to iterate on one image without changing everything, so this is pretty cool
  • anyi09881 11 days ago
    Curious what's the actual cost for each edit? Will this infra always be reliable?
    • CamperBob2 11 days ago
      I was able to clone the repo and run it locally, even on a Windows machine, with only minimal Python dependency grief. Takes about a minute to create or edit an image on a 4090.

      It's pretty impressive so far. Image quality isn't mind-blowing, but the multi-modal aspects are almost disturbingly powerful.

      Not a lot of guardrails, either.

      • hyuuu 7 days ago
        could you elaborate on the multi modal aspect of this model?
  • jay_1999 10 days ago
    [dead]
  • sunny-sunny 11 days ago
    [dead]
  • 545999961 10 days ago
    [dead]
  • convolution8 8 days ago
    [dead]
  • toast12345 8 days ago
    [flagged]
  • toast12345 8 days ago
    [flagged]
  • kazishariar 11 days ago
    [flagged]
    • illumanaughty 11 days ago
      We've been manipulating photos as long as we've been taking them.
    • handfuloflight 11 days ago
      Art is what you can get away with. (Andy Warhol)